Академический Документы
Профессиональный Документы
Культура Документы
doi:10.1038/nature13586
the signal single photon counts; the idlers are not detected. No coincidence
detection is required.
The peculiar feature of this interferometer is that no detected photon
has taken path d. Yet, in our experiment, it is precisely here where we put
the object to be imaged. The key to this experiment is how the signalsource information carried by the undetected idler photon depends on
T. For,
if T5 0, an idler detected after D3, coincident with a signal count
at gis or his , would imply the signal source was NL2. Detection of a
signal photon without a coincident idler would imply the signal source
was NL1. This which-source information destroys interference because
it makes the quantum states overlapping at BS2 distinguishable. If
T 5 1, the idler photon carries no which-source information. The signal states overlapped at each output of BS2 are then indistinguishable;
thus the interference term in equation (2) appears. The above arguments
are valid even though the idler photons are not detected, for it is only
the possibility of obtaining which-source information that matters in
this experiment.
Our experiment has a connection to interaction-free measurements11,12.
Note that Ph ~0 if no object is placed in the set-up (T 5 1 and c 5 0).
Now insert an opaque object (T 5 0) so thatPh . 0, and monitor the
idler reflection from D3. Coincident counts in his and the idler detector
reveal that the object is present even though no photon interacted with
the object. With our set-up it is thus possible to realize non-degenerate
interaction-free imaging.
With O and D2 removed,
equation
(1) would be an ordinary two
particle entanglement13, cis dii zeis f ii . With them in, dii ?Teic f ii ,
which creates equation (1). A normal two-particle entanglement has
changed into an interesting single-particle superposition, which is especially rich when T and c are transverse-position dependent.
We expand the conceptual arrangement of Fig. 1 into an imaging
system (Fig. 2). We replace the photon counters with cameras sensitive
to single photons and the uniform object with one bearing features, that
is, T 5 T(x, y) and c 5 c(x, y) depend on transverse position (x, y). Our
source produces spatially entangled photon pairs14,15. Sharp spatial correlations between signal and idler in the object plane and confocal lens
systems16 (see Methods) guarantee a point-by-point correspondence between the object plane and the detector surface on the camera.
The intensity image (non-constant transmittance) is due to transverseposition-dependent which-source information carried by the undetected
idler photons. The phase image is of a different nature: it is due to the fact
that the position-dependent phase shift
on the
idler
inpathic d is
photons
17
ic
actually
passed to the signal; that is , cis Te f ii z eis f ii ~ Te cis
zeis f ii . Remarkably, the idler beam f ii alone does not even carry the
phase pattern, and without detection in coincidence it could not be used
to obtain the phase image18,19.
We will now show images obtained by detecting 810-nm photons
with a camera capable of single-photon sensitivity at this wavelength,
when three different objects are illuminated by 1,550-nm photons, to
which our camera is blind (see Methods). First, a cardboard cut-out
placed into the path D1D2 is imaged. Next, we show that a positiondependent phase shift produces an image even when the object is opaque
(an etched silicon plate) or invisible (etched silica plate) at the detection
Institute for Quantum Optics and Quantum Information, Austrian Academy of Sciences, Boltzmanngasse 3, Vienna A-1090, Austria. 2Vienna Center for Quantum Science and Technology (VCQ), Faculty of
Physics, University of Vienna, A-1090 Vienna, Austria. 3Quantum Optics, Quantum Nanophysics, Quantum Information, University of Vienna, Boltzmanngasse 5, Vienna A-1090, Austria. {Present address:
Cornell University, 159 Clark Hall, 142 Science Drive, Ithaca, New York 14853, USA.
2 8 AU G U S T 2 0 1 4 | VO L 5 1 2 | N AT U R E | 4 0 9
RESEARCH LETTER
BS1
D1
NL1
300
800
c
Pixels
600
200
400
100
200
O
0
100
200
300
Pixels
D3
BS2
f
e
NL2
D1
NL1
200
800
100
400
0
0
100 200
Pixels
300
900
200
600
100
300
300
100 200
Pixels
2,500
200
Pixels
2,000
1,500
100
1,000
L4
500
532 nm
0
L3
1
M1
L3
100
200
300
500
Pixels
400
200
0
BS
D5
L2 D2
400
z (nm)
2
L2
L4
D4
HWP
300
Counts
PBS
1,200
Counts
600
d
300
Counts
WPs
500
Pixels
D2
400
Pixels
L1
3 mm
Counts
NL2
L5
L6
4
3
Filters
3
2
1
1
mm
y(
EMCCD
0
0
m)
x (m
4 1 0 | N AT U R E | VO L 5 1 2 | 2 8 AU G U S T 2 0 1 4
LETTER RESEARCH
a
400
400
300
200
200
100
100
0
400
400
300
300
200
200
100
100
0
0
100
200
300
400
500
600
700
800
Counts
Pixels
Counts
Pixels
300
Pixels
b
2
z (m)
1
3
y
2
(m
1
0
m
x (m
METHODS SUMMARY
A detailed schematic of our imaging set-up is shown in Fig. 2. A 532-nm linearly
polarized Gaussian pump laser beam focused by lens L1 on plane 1 is divided at a
polarizing beam splitter (PBS) and coherently illuminates two identical periodically
poled potassium titanyl phosphate (ppKTP) crystals, NL1 and NL2. The PBS plus
wave plates (WPs) are used to control the relative amplitudes and phases between the
reflected and transmitted pump beams. With an extra half-wave plate (HWP) in the
reflected beam, both beams have the same polarization. The 1,550-nm idler amplitude produced at NL1 is reflected by dichroic mirror D1, through which the 810-nm
signal and the pump are transmitted. Dichroic mirror D4 transmits 532-nm light
and reflects 810-nm light. A long-pass filter (not shown in the figure) placed directly
before the object O blocks any residual 532-nm or 810-nm light. The 1,550-nm
amplitude from NL1 illuminates the object O and is then overlapped with the pump
beam at dichroic mirror D2 that transmits 532-nm light and reflects 1,550-nm light.
Lens pairs L2L29, L3L39, and L4L49 image plane 1 onto plane 3, thereby ensuring that pump, idler and signal, respectively, are identical in these planes, thus contributing to obtain high interference visibility21 (see Methods). Lenses L5 and L6
together with L39 and L49 image object plane 2 onto the camera surface.
The 810 6 1.5 nm photons are detected (without heralding) in both outputs of the
BS using an EMCCD camera that exhibits single-photon sensitivity at 810 nm, but
has a negligible response at 1,550 nm.
Online Content Methods, along with any additional Extended Data display items
and Source Data, are available in the online version of the paper; references unique
to these sections appear only in the online paper.
Received 26 January; accepted 11 June 2014.
1.
Figure 5 | Phase imaging of a 2p step at 820 nm. a, The top picture was taken
with the object (shown in b) placed in the 820-nm beam between L4 and L49;
in the bottom picture, the object was placed in the 1,515-nm beam in path
D1D2. b, Three-dimensional rendering of the design overlaid with stylus
profilometer scans (blue dots) of the actual etch depth.
2.
3.
4.
5.
and 1,515 nm (see Methods). The object (Fig. 5b) has an etch depth of
1,803 nm, imparting a relative phase shift of ,2p for 820-nm light. Thus
the object is invisible when placed in the path of the detected photons
between L4 and L49 (top of Fig. 5a). This same etch depth gives an ,p
phase step for 1,515-nm light, so when this same object is placed in the
path D1D2 of undetected photons, an image seen in the contrast of
constructive to destructive interference is retrieved in the 820-nm outputs (bottom of Fig. 5a).
In summary, we have presented a quantum system for intensity and
phase imaging where the photons that illuminate the object are not
detected and the photons that are detected do not illuminate the object.
We image objects that are either opaque or invisible at the detection
wavelength (near-infrared) by illuminating three different objects with a
wavelength to which our detector is blind. This experiment is fundamentally different to ghost imaging69 as it relies on single-photon interference and does not require coincidence detection. Furthermore, our
technique could be used for non-degenerate interaction-free imaging,
with potential applications spanning biological imaging to the inspection of integrated circuits. Our system can realize grey-scale intensity
or phase imaging, and it can be modified in order to measure spectral
features (spectral imaging)20.
We have demonstrated that our technique does not require the laser or
the detector to function at the same wavelength as that of the light probing
the object. Additionally, any nonlinear process can be used as a source, and
this provides flexibility in the wavelength range for both detection and
illumination of the object. In particular, in spontaneous parametric downconversion (as used here), the only absolute restriction is that the sum of
the two photon energies equals that of the pump photons. We have shown
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
Feynman, R. P., Leighton, R. B. & Sands, M. The Feynman Lectures on Physics Vol. III,
Chs 1 and 3 (Addison-Wesley, 1964).
Mandel, L. Coherence and indistinguishability. Opt. Lett. 16, 18821883 (1991).
Zou, X. Y., Wang, L. J. & Mandel, L. Induced coherence and indistinguishability in
optical interference. Phys. Rev. Lett. 67, 318321 (1991).
Wang, L. J., Zou, X. Y. & Mandel, L. Induced coherence without induced emission.
Phys. Rev. A 44, 46144622 (1991).
White, A. G., Mitchell, J. R., Nairz, O. & Kwiat, P. G. Interaction-free imaging. Phys.
Rev. A 58, 605613 (1998).
Abouraddy, A. F., Stone, P. R., Sergienko, A. V., Saleh, B. E. A. & Teich, M. C.
Entangled-photon imaging of a pure phase object. Phys. Rev. Lett. 93, 213903
(2004).
Gatti, A., Brambilla, E. & Lugiato, L. Quantum imaging. Prog. Opt. 51, 251348
(2008).
Pittman, T. B. et al. Two-photon geometric optics. Phys. Rev. A 53, 28042815
(1996).
Aspden, R. S., Tasca, D. S., Boyd, R. W. & Padgett, M. J. EPR-based ghost imaging
using a single-photon-sensitive camera. New J. Phys. 15, 073032 (2013).
Wiseman, H. M. & Mlmer, K. Induced coherence with and without induced
emission. Phys. Lett. A 270, 245248 (2000).
Elitzur, A. C. & Vaidman, L. Quantum mechanical interaction-free measurements.
Found. Phys. 23, 987997 (1993).
Kwiat, P., Weinfurter, H., Herzog, T., Zeilinger, A. & Kasevich, M. A. Interaction-free
measurement. Phys. Rev. Lett. 74, 47634766 (1995).
Horne, M. in Experimental Metaphysics Vol. 1. (eds Cohen, R. S., Horne, M. & Stachel,
J.) 109119 (Kluwer Academic, 1997).
Howell, J. C., Bennink, R. S., Bentley, S. J. & Boyd, R. W. Realization of the EinsteinPodolsky-Rosen paradox using momentum-and position-entangled photons
from spontaneous parametric down conversion. Phys. Rev. Lett. 92, 210403
(2004).
Walborn, S. P., Monken, C. H., Padua, S. & Souto Ribeiro, P. H. Spatial correlations in
parametric down-conversion. Phys. Rep. 495, 87139 (2010).
Tasca, D. S., Walborn, S. P., Souto Ribeiro, P. H., Toscano, F. & Pellat-Finet, P.
Propagation of transverse intensity correlations of a two-photon state. Phys. Rev. A
79, 033801 (2009).
Horne, M. A., Shimony, A. & Zeilinger, A. Two particle interferometry. Phys. Rev. Lett.
62, 22092212 (1989); Two particle interferometry. Nature 347, 429430
(1990).
Ribeiro, P. H. S., Padua, S., Machado da Silva, J. C. & Barbosa, G. A. Controlling the
degree of visibility of Youngs fringes with photon coincidence measurements.
Phys. Rev. A 49, 41764179 (1994).
Abouraddy, A. F., Stone, P. R., Sergienko, A. V., Saleh, B. E. A. & Teich, M. C.
Entangled-photon imaging of a pure phase object. Phys. Rev. Lett. 93, 213903
(2004).
Zou, X. Y., Grayson, T. P. & Mandel, L. Observation of quantum interference effects
in the frequency domain. Phys. Rev. Lett. 69, 30413044 (1992).
2 8 AU G U S T 2 0 1 4 | VO L 5 1 2 | N AT U R E | 4 1 1
RESEARCH LETTER
21. Grayson, T. P. & Barbosa, G. A. Spatial properties of spontaneous parametric
down-conversion and their effect on induced coherence without induced emission.
Phys. Rev. A 49, 29482961 (1994).
Acknowledgements We thank M. Horne for reading the manuscript, clarifying
suggestions and many discussions, P. Enigl for designing the figures for the objects,
D. Greenberger and S. von Egan-Krieger for discussions, and C. Schaeff for equipment
loans. Microfabrication was carried out at the Center for Micro- and Nanostructures
(ZMNS) of the Vienna University of Technology. We acknowledge D. Ristanic for
assistance with cryogenic Si etching and M. Schinnerl for contact mask production.
AW) through a fellowship
G.B.L. was funded by the Austrian Academy of Sciences (O
from the Vienna Center for Science and Technology (VCQ). S.R. is funded by an EU
Marie Curie Fellowship (PIOF-GA-2012-329851). This project was supported by OAW,
the European Research Council (ERC Advanced grant no. 227844 QIT4QAD, and SIQS
grant no. 600645 EU-FP7-ICT), and the Austrian Science Fund (FWF) with SFB F40
(FOQUS) and W1210-2 (CoQus).
Author Contributions A.Z. initiated this research. G.B.L., V.B., R.L., S.R. and A.Z.
designed the experiment. G.B.L., V.B. and R.L. carried out the experiment. G.D.C.
fabricated the silicon and silica phase masks. All authors contributed to the writing of
the manuscript.
Author Information Reprints and permissions information is available at
www.nature.com/reprints. The authors declare no competing financial interests.
Readers are welcome to comment on the online version of the paper.
Correspondence and requests for materials should be addressed to
G.B.L. (gabriela.barreto.lemos@univie.ac.at) and A.Z.
(anton.zeilinger@univie.ac.at).
4 1 2 | N AT U R E | VO L 5 1 2 | 2 8 AU G U S T 2 0 1 4
LETTER RESEARCH
METHODS
Down-conversion sources. The 532-nm pump beam is generated by a frequencydoubled diode-pumped solid-state laser (Coherent Sapphire SF) and is focused
onto the two periodically poled potassium titanyl phosphate (ppKTP) crystals with
dimension 1 mm 3 2 mm 3 2 mm and poling period 9.675 mm for type-0 phase
matching. The crystals are spatially oriented so down-conversion occurs when the
CW pump beam is horizontally polarized (both the signal and idler produced are
also horizontally polarized). In order to conform to the phase-matching conditions
for 810-nm and 1,550-nm photons, NL1 (NL2) is heated to 83.7 uC (84.7 uC). When
the set-up is adjusted to produce 820-nm and 1,515-nm photons (to be used with the
fused silica phase object), NL1 (NL2) is heated to 39.2 uC (39.7 uC). All images were
obtained with 150-mW pump power.
Wavelength filtering. Inside the interferometer, D1 is used to separate the 810-nm
photons from the 1,550-nm photons. Mirror D1 (and also D2) reflects about 93% of
1,550-nm light and transmits about 99% of 810-nm light. Most of the pump beam
going through NL1 is transmitted through both D1 and D4 (each with a transmittance of around 97% at 532 nm) and therefore almost never reaches BS. The dichroic
mirror D5 additionally transmits some 532-nm light (around 25%), so some of the
pump beam that goes through NL2 as well as some of the remaining pump beam
from NL1 are discarded there. All remaining pump beam light is eliminated with
either filters or the imaging object itself. The silicon sample is opaque to both 532-nm
and 810-nm light, thus completely blocking these wavelengths along the path D1
D2. When the other samples are used, a long-pass filter is placed just before the object
to cut out these lower wavelengths. The remaining 532-nm light that is not separated
out through the dichroic mirrors or object is blocked in front of the camera by three
filters. A 3-nm narrowband filter centred at 810 nm and two long pass filters were
attached directly to the front of the camera. As it utilizes a silicon-based detector, the
camera (Andor Luca-R EMCCD) does not detect 1,550-nm photons. Nonetheless, a
combination of spectral filters guarantees that neither 1,550-nm photons nor 532-nm
pump photons reach the camera.
Imaging lens systems. As it is crucial that the down-converted photons be identical,
we use confocal lens systems to image plane 1 onto plane 3 (see Fig. 2), thus ensuring
that the pump beams at NL1 and NL2 are identical, the 810-nm photons when they
combine at the BS are identical, and the 1,550-nm photons are identical from NL2
onward. Lenses L2 and L29 image plane 1 of the pump onto plane 3, and similarly L3
(L4) and L39 (L49) image plane 1 onto plane 3 for the 1,550-nm (810-nm) photons.
Lenses L5 and L6 in combination with L49 image plane 2 onto the EMCCD camera.
Lenses L2, L29, L3, L39, L4, L49 have a focal length of F1 5 75 mm. The distance from
plane 1 to each of L2, L3 and L4 is 75 mm; from those lenses to plane 2 is another
75 mm; from plane 2 to L29, L39 and L49 is also 75 mm; and from those lenses to plane 3
is yet another 75 mm. This ensures that the photons produced in both crystals have
the same waist and divergence when they reach the BS. Lenses L5 and L6 have a focal
length of F2 5 150 mm. They are placed 150 mm after plane 3 and 150 mm before the
camera. The total imaging magnification from the object to the camera is given by
F2 ls
, where ls and li are the wavelengths of the signal and idler photons, respectively.
F1 li
Optical path lengths. In our single photon interferometer the paths D1D4BS
and D1D2BS need to be equal, even though no detected photons actually follow
the entire path D1D2BS. To assure indistinguishability of the emission in the two
crystals (NL1 and NL2) the time delay between the arrival of the signal and idler
for each of the two crystals must be the same. The path length difference between
the signal and idler for the pair from NL1 is the distance NL1D1D2BS subtracted
from the distance NL1D1D4BS. The path length difference between the signal
and idler for the pair from NL2 is zero since the down-conversion is collinear. Thus,
we see that the optical path lengths between D1D4BS and D1D2BS must be
equal to within the coherence length of the photons. The coherence length of the
photons is in our case determined by the filtering (3 nm), so we approximate the
coherence length to be 0.2 mm. The other relevant optical path lengths are the paths
PBSD1D2NL2 and PBSM1NL2. The differential distance between these paths
must be within the coherence length of the laser, which in our case is approximately
200 m.
Intensity object. Our intensity object is constructed from 0.33-mm-thick card stock
with images defined by laser cutting. The images on the object were each 3 mm high.
Microfabricated silicon phase sample. The first custom phase sample consists of
500-mm-thick double-side polished (100)-oriented single-crystal silicon with imaging
targets defined on one face using standard microfabrication techniques. The absorption coefficient of silicon is ,1,000 cm21 at 810 nm (ref. 22), and it is ,1024 cm21
at 1,550 nm (ref. 23). Processing begins by cleaving a 75-mm-diameter silicon wafer
to obtain chips with lateral dimensions of 25 mm 3 25 mm. The cleaved chips are
patterned using conventional optical contact lithography followed by plasma etching. In order to generate a relative p-phase shift at 1,550 nm, features are etched to
a depth of approximately 310 nm (nominal height of 321 nm using a refractive index
of silicon of 3.48; ref. 24) into the exposed Si surface using a cryogenic (2108 uC) SF6/
O2 reactive-ion etching (RIE) process protected with a positive photoresist mask. To
improve thermal transfer, the silicon chips are mounted on a carrier wafer using a
thin layer of vacuum grease. Additionally, in order to minimize variations in the
overall etch depth and thus resulting phase shift from the imaging targets, the feature
linewidth is kept constant over the lithographic pattern to mitigate the effects of
aspect-ratio dependent etching (or RIE lag). After etching, the chips are removed
from the carrier wafer and the masking resist and mounting film are stripped using
a combination of organic solvents and oxygen plasma ashing. In order to eliminate
spurious reflections from the polished surfaces, a dual-sided silicon nitride antireflection (AR) coating is deposited via plasma-enhanced chemical vapour deposition (PECVD) using He-diluted SiH4 and NH3 as reactive process gases. The deposition process yields quarter-wave optical thickness layers at a target film thickness of
(with a refractive index of 1.9 at the imaging wavelength of 1,550 nm).
2,040 A
In order to achieve the highest contrast, the relative path-length difference between
the etched and non-etched regions should be equal to a half wavelength of 1,550-nm
light adjusted for the difference in the indices of refraction of silicon and air. This gives
a target thickness difference of 321 nm (for a refractive index of silicon of 3.48). Given
the slight error in etch depth, the actual thickness difference is 310 nm, which is still
sufficient to obtain high contrast images.
Microfabricated fused silica phase sample. Similar to the silicon phase object described above, the fused silica phase sample, cleaved from a 500-mm-thick glass wafer,
is constructed via a standard lithographic and reactive ion etching process. In this
case the same mask pattern is once again defined with contact lithography. In order
to transfer the features into the fused silica, a high-power inductively coupled plasma
(ICP) RIE process is required (150 W ICP, 250 W RF powers) with an etch chemistry
consisting of SF6 and Ar. Given the poor selectivity to the masking resist, a thick
(10 mm) coating of AZP4620 photoresist is required. The target etch depth of
1,788 nm is achieved within roughly 10 min at room temperature. Given the high
plasma energy, thermalization with the cooled carrier wafer is key. Due to nonuniformities in thermal contact with the carrier, we observe significant variation in
etch depth (6200 nm) across the surface of the 25 mm 3 25 mm pattern. No AR
coating is employed given the small Fresnel reflection (4%) from the low-index silica
substrate.
For 820-nm light, an exact 2p phase shift is given by a thickness difference of
1,811 nm (using an index of refraction of 1.45)25; after processing, the average etch
depth recorded for the fused silica sample is 1,803 nm.
Interference visibility. In order to quantify the visibility in our imaging experiment,
we detect the total intensity of 810-nm photons at one output of BS as a function of
the relative phase between the pump beams that illuminate each crystal. Extended
Data Fig. 1 shows a plot of the count rate measured with an avalanche photodiode
when no object is present. The red circles show the experimental points, and the best
fitting sinusoidal function (red line) gives a visibility of (77 6 1)%. The visibility for
our experiment is given not only by losses in both the 1,550-nm and 810-nm arms of
the interferometer, but also by residual imperfections in the alignment for the two
idler beams. The blue squares correspond to data obtained when the path NL1NL2
is completely blocked, which results in zero interference visibility. Interference only
arises if the idler between the two crystals is unblocked, for only then is its source, and
therefore also the source of its signal sister, unknowable.
Showing that induced emission is negligible in the experiment. In order to
demonstrate in our experiment that the 1,550-nm photons from NL1 do not induce
down-conversion in NL2, we show in Extended Data Fig. 2 the count rates for 810-nm
photons originating at NL2 when the 1,550-nm beam between D1 and D2 was blocked
(blue crosses) and unblocked (red dots). The mean count rate and the standard
deviation were obtained by analysing data obtained over 40 s. The blue diamonds
show that the ratio of the count rates for the blocked and unblocked configuration is
very close to 1 irrespective of the pump power.
22.
23.
24.
25.
Jellison, G. E. Jr. & Modine, F. A. Optical absorption of silicon between 1.6 and 4.7
eV at elevated temperatures. Appl. Phys. Lett. 41, 180 (1982).
Khalaidovski, A., Steinlechner, J. & Schnabel, R. Indication for dominating surface
absorption in crystalline silicon test masses at 1550 nm. Class. Quantum Grav. 30,
165001 (2013).
Malitson, I. H. Interspecimen comparison of the refractive index of fused silica.
J. Opt. Soc. Am. 55, 1250 (1965).
Bass, M. Handbook of Optics Vol. 2, 2nd edn (Optical Society of America, 1995).
RESEARCH LETTER
Extended Data Figure 1 | Visibility of the experiment. The count rates were
recorded with the path D1D2 both unblocked (red dots) and blocked (blue
squares) as the relative phase between the transmitted and reflected beams of
the PBS was varied. The red line is a sine curve fit for the experimental data
giving (77 6 1)% visibility. The error bars are smaller than the size of the data
points.
LETTER RESEARCH
diamonds show the ratio of the count rates for the blocked and unblocked
configuration. The linear fit for this data (black line) gives an angular coefficient
of (2 6 4) 3 1025 (mW)21.